Clinical Features of 24 Patients With Rebound‐Associated Vertebral Fractures After Denosumab Discontinuation: Systematic Review and Additional Cases

ABSTRACT

 

We aimed to study the clinical and imaging characteristics of patients sustaining vertebral fractures after denosumab discontinuation. For this purpose, we conducted a computerized advanced literature search that identified 13 published cases, and we additionally included another 11 new cases from our centers. Twenty‐four postmenopausal women with vertebral fracture(s) after denosumab discontinuation, experiencing 112 fractures in total, were analyzed. The mean number of fractures per patient was 4.7. The most commonly affected vertebrae were T12 and L1. All fractures occurred 8 to 16 months after the last denosumab injection. Eighty‐three percent of the patients were treatment naïve, whereas 33% had prevalent vertebral fractures. Five (23%) patients were on concurrent aromatase inhibitor treatment. When patients were divided according to treatment duration with an arbitrary cut‐off of 2 years, those with ≤2 years of denosumab treatment had fewer fractures compared with those with >2 years (mean ± SEM fractures 3.2 ± 0.7 versus 5.2 ± 1.4, p = 0.055). Vertebroplasty was used in 5 patients, resulting in additional clinical vertebral fractures in all cases. We conclude that vertebral fracture(s) after denosumab discontinuation are in the majority of patients multiples, and they occur a few months after the effect of the last dose is depleted. Therefore, patients should not delay or omit denosumab doses. Fractures are typically osteoporotic, located at the lower thoracic and the upper lumbar spine. Vertebroplasty is an unsuccessful treatment strategy for such patients. © 2017 American Society for Bone and Mineral Research.

DENOSUMAB, DISCONTINUATION, FRACTURE, OSTEOPOROSIS, VERTEBRAL

Introduction

Denosumab, a monoclonal antibody against the receptor activator of nuclear factor κ‐Β ligand (RANKL), is a potent antiresorptive agent⁽¹⁾ commonly prescribed in patients with postmenopausal osteoporosis. Discontinuation of denosumab results in a rebound response of bone turnover markers, which rise above baseline at 3 months and remain elevated until reaching again baseline levels approximately 30 months after the last dose. Bone mineral density (BMD) gains are also lost, and BMD values reach baseline values after 2 years off‐treatment.⁽², ³⁾ Furthermore, recent case reports have raised concerns regarding increased vertebral fragility after denosumab discontinuation⁽⁴, ⁵, ⁶, ⁷, ⁸⁾ and lead to the notion that patients should not stop denosumab regardless of BMD values.⁽⁹⁾ Published cases report clinical fractures with an acute onset of symptoms, which are most commonly multiple. However, and despite the fact that the increase in bone turnover and decrease in BMD are universal after denosumab discontinuation, only a minority of patients sustain such fractures. The exact pathophysiology behind this increased fracture risk is unknown; the main mechanism proposed includes excessive remodeling rate, which is higher than pretreatment⁽⁴, ⁵, ⁶, ⁷, ⁸⁾ and is expected to affect the trabecular bone more and faster than the cortical bone, a hypothesis that could explain the multiple vertebral fractures reported in these cases.^(4–8) However, to date no risk factors have been identified that could predict with certainty these multiple vertebral fractures.

The primary aim of the present systematic review was to identify those clinical or imaging characteristics that could be associated with increased risk of vertebral fractures upon denosumab discontinuation, observations potentially useful for the clinical management of patients for whom discontinuation may be considered.

Materials and Methods

Literature search

Computerized advanced search for primary evidence was performed in PubMed (Public/Publisher MEDLINE) electronic database on November 30, 2016, by the combination of terminological (MeSH terms) and methodological search filters, as elsewhere reported.⁽¹⁰⁾ Search was not limited by publication time and not restricted to English literature. The original detailed query was: (“denosumab”[MeSH Terms] OR “denosumab”[All Fields]) AND (discontinuation[All Fields] OR stop[All Fields]) AND (“fractures, bone”[MeSH Terms] OR (“fractures”[All Fields] AND “bone”[All Fields]) OR “bone fractures”[All Fields] OR “fracture”[All Fields]).

Afterward, the bibliographic search was extended to the “Related Articles” link next to each selected article in PubMed and its references. Finally, automatic alerts were activated in PubMed (“My NCBI”) to add relevant articles published after the initial search (last update December 22, 2016).

New cases

Besides the cases already reported in the literature, the authors additionally included in the analysis new patients with vertebral fractures after denosumab discontinuation that had been referred to their centers in the meantime.

Cases description

Parameters obtained for each patient include: demographic data, duration of denosumab treatment, time from last denosumab injection to fractures, site of fractures, presence of previous vertebral/non‐vertebral fractures, lumbar spine DXA T‐score at denosumab initiation and discontinuation, the reason for denosumab discontinuation and the post‐fracture management of the patients. The diagnosis of the vertebral fractures was based on lateral lumbar spine X‐rays or vertebral fracture assessment (VFA) and was confirmed by dorsolumbar magnetic resonance imaging (MRI).

Results

Thirty‐one articles were initially identified. No additional study was added after the “hand‐searching.” After the screening of the articles, 5 articles were considered eligible for this systematic review.⁽⁴, ⁵, ⁶, ⁷, ⁸⁾ The 3 cases described by Aubry‐Rozier and colleagues⁽⁵⁾ were all included in the 9 cases described by Lamy and colleagues.⁽⁸⁾ We did not include the cases of fracture reported in clinical trials or their post hoc analyses⁽¹¹⁾ because we could not have access to patients’ details. Twenty‐four postmenopausal women with vertebral fracture(s) after denosumab discontinuation (13 already reported in the literature plus 11 new cases from our centers) were retrieved, experiencing 112 fractures in total. We noticed no difference between already reported and new cases. None of the cases experienced any non‐vertebral fracture during the follow‐up period. All cases are summarized in Table 1.

Clinical and Imaging Characteristics of All Women Reported in the Literature to Have Sustained Vertebral Fractures After Denosumab Discontinuation

Reference	Age (years) at VFxs	Time on Dmab (Yrs)	Last injection to VFx (months)	No. of VFxs	Site of VFxs	Prevalent VFxs	Prevalent non‐VFxs	T‐score_LS at Dmab initiation	T‐score_LS at Dmab stop	Reason for Dmab discontinuation	Post VFxs management	Comments
Popp, Osteoporos Int 2016(6)	53	3	9	7	T8, T10, T12, L1–4	0	0	–2.0	–0.8	End of AI and normalization of BMD	NR	Breast cancer under AI
Polyzos, Endocrine 2016(7)	62	2	14	1	L3	T8	0	–2.8	–0.2	Normalization of BMD	Calcium/vitamin D–intention for TPTD	Secondary hyperparathyroidism
Polyzos, Endocrine 2016(7)	61	1	12	2	T12, L1	0	0	NR	NR	Became osteopenic	Dmab	Pretreated 1year with SR and 5 years with raloxifene
Anastasilakis, Osteoporos Int 2016(4)	55	3	8	3	T12, L1, L3	0	0	–2.5	–1.8	Became osteopenic	TPTD	—
Lamy, JCEM 2016(8)and Aybrie‐Rozier, Osteoporos Int 2016(5)	55	2.5	9	5	T11, T12, L2–L4	0	0	–3.1	–2.3	Became osteopenic	TPTD	—
Lamy, JCEM 2016(8)	56	4	10	9 (7 + 2)	T7, T8, T10–12, L1, L2, L4, L5	0	0	–2.8	–2.2	Tx duration	Vertebroplasty – Dmab + TPTD	—
Lamy, JCEM 2016(8) and Aybrie‐Rozier, Osteoporos Int 2016(5)	59	3.5	10	2	T11, T12	0	0	–3.1	–2.4	Became osteopenic	TPTD	Rheumatoid arthritis/never on glucocorticoids
Lamy, JCEM 2016(8)	58	1	11	8	T8–T11, L1, L3–L5 (T11 and L1: deterioration)	T11, L1	0	–3.9	–3.5	Patient's wish	TPTD	—
Lamy, JCEM 2016(8)	63	1	12	1	T10	0	0	–1.7	NR	Tx omission	Dmab	Breast cancer under AI
Lamy, JCEM 2016(8)	65	4	10	6	T5, T8, T12, L2–L4 (T12: deterioration)	T12	0	–3	–2.3	Tx duration	Dmab + TPTD	—
Lamy, JCEM 2016(8)	73	1	11	5	T12, L2–L5	L1	Hip	–4.5	–3.1	Patient's wish	No Rx	Bisphosphonate for 3 years, 11 years before Dmab initiation
Lamy, JCEM 2016(8)0	81	3	16	5 (3 + 2)	T9, T12, L1–L3	0	0	–3.9	–3.1	End of AI	Vertebroplasty – ZOL	Breast cancer under AI
Lamy, JCEM 2016(8) and Aybrie‐Rozier, Osteoporos Int 2016(5)	80	2.5	16	9 (3 + 6)	T5–9, T11–L2	0	0	–4.1	–3.7	Patient's wish	Vertebroplasty – No Rx	—
New case	60	3.5	12	7	T5, T11, L1–L5	T7, T10, T12	0	–2.3	–2.1	Dental Tx	Dmab + TPTD	Glucocorticoids (inflammatory disease). Breast cancer under AI
New case	65	5	11	5	T8, T11–L2	0	0	–3.4	–2.6	Tx duration	ZOL + TPTD
New case	62	5	11	5 (4 + 1)	T10–L2	0	0	–4.1	–2.8	Tx omission	Vertebroplasty – Dmab + TPTD	ALN before Dmab (short time, adverse effect)
New case	48	4.5	10	5	T4, T8, T9, T12, L2	0	0	–1.9	–1.3	End of AI and normalization of BMD	Dmab	Breast cancer under AI
New case	83	3	10	2	T12, L3	0	0	–	–	Patient's negligence	TPTD	Pretreated 1 year with TPTD
New case	82	2	16	2	T11, L1	L3, L5	0	–2.6	–2.0	Became osteopenic	TPTD	—
New case	68	3.5	8	2	T12, L1	0	0	–3.4	–2.4	Became osteopenic	ZOL	—
New case	71	2.5	10	4	Τ9, L1, L2, L5	T6, T7, T12, L3, L4	0	–2.6	–2.1	Became osteopenic	TPTD	—
New case	63	5	10	5	T1, T11, L1 L3, L4	0	0	–1.2	–0.5	End of AI and normalization of BMD	Dmab	Breast cancer under AI
New case	61	3	9	7 (3 + 4)	T6–8, T11–12, L3, L4	L2	0	–2.5	–1.5	Became osteopenic	Vertebroplasty + ZOL	—
New case	54	4	13	5	T11, T12, L3–5	0	0	–2.3	–1.6	Became osteopenic	No Rx	—
Mean (range)	64.1 (48–83)	2.9 (1–5)	11.2 (8–16)	4.7 (1–9)	–	8/24	1/24	–2.9 [(–1.2)–(–4.5)]	–2.1 [(–0.2)–(–3.70)]		TPTD: 8 Dmab: 4 ZOL: 3 TPTD + Dmab: 4 TPTD + ZOL: 1

AI = aromatase inhibitor; ALN = alendronate; BMD = bone mineral density; Dmab = denosumab; FN = femoral neck; L = lumbar; LS = lumbar spine; meno = age at menopause; NR = not reported; Rx = treatment; SR = strontium ranelate; T = thoracic; TPTD = teriparatide; Tx = treatment; VFx = vertebral fracture; ZOL = zoledronic acid.

The number of fractures per patient was 4.7 (mean) / 5.0 (median) with a range from 1 to 9. Fracture location was as depicted in Fig. 1. Ninety‐two percent of the patients had more than one vertebral fracture. Vertebra T12 was the most commonly affected (17/24) followed by L1 (14/24), L3 (13/24), T11 (12/24), and L2 (12/24) (Fig. 1). When patients were divided according to treatment duration with an arbitrary cut‐off of 2 years, those with ≤2 years of denosumab treatment had fewer fractures compared with those with >2 years (mean ± SEM fractures 3.2 ± 0.7 versus 5.2 ± 1.4, p = 0.055). There was no significant difference between the two subgroups for age or lumbar spine BMD at denosumab initiation or discontinuation. The only difference concerned the prevalent vertebral fractures (66.7% in those treated ≤2 years versus 22.2% in those treated >2 years), which is probably coincidental. A broad biological and radiological assessment at the time of vertebral fractures excluded a secondary cause of bone fragility. The most frequent reasons for discontinuing denosumab were to have reached the osteopenic or normal BMD range, treatment duration, or the end of treatment with aromatase inhibitors (AI) (Table 1).

The distribution of fractures among vertebrae in the sum of the patients (n = 24).

Twenty of the 24 patients (83%) were treatment naïve. The remaining 4 had received previous treatment for osteoporosis: strontium ranelate and raloxifene (n = 1), teriparatide (n = 1), and bisphosphonates (n = 2) (Table 1). Eight patients (33%) had prevalent vertebral fractures, while 1 of them had also a non‐vertebral fracture. Only 1 patient had received glucocorticoids. Five patients (21%) were receiving AI for breast cancer. Seventy‐five percent of the patients had a lumbar spine T‐score ≤ –2.5 SD at the time of denosumab discontinuation.

After the incidence of fractures, 5 patients were subjected to vertebroplasty, all unsuccessful because of the occurrence of new vertebral fractures. Teriparatide was the most commonly administered pharmaceutical agent after the fractures, followed by the combination of teriparatide and denosumab or denosumab reinitiation alone (Table 1).

Discussion

Denosumab discontinuation has been linked to increased lumbar spine fragility in a small percentage of patients.⁽⁹⁾ The exact mechanisms leading to vertebral fracture(s) in these cases are yet unknown, and therefore an official position regarding the appropriate strategy to prevent or treat these events is lacking up to now, although it has been suggested that a long‐acting bisphosphonate would be appropriate if/when denosumab therapy is stopped.⁽⁹⁾ We aimed to identify the clinical and imaging characteristics of these patients in an attempt to both study and understand this phenomenon and to retrieve specific characteristics that could help at the identification of patients at increased fracture risk after denosumab discontinuation.

Several mechanisms have been proposed to explain the increased fragility of a subset of patients after denosumab discontinuation. The most prevailing hypothesis is that of markedly increased bone turnover. It has been speculated that the rebound effect is more prominent as the number of denosumab doses increases.⁽¹²⁾ Thus, discontinuation after long‐term therapy may result in an even greater risk than discontinuation after only 2 years.⁽¹²⁾ We also found that patients treated with denosumab for more than 2 years had more fractures than those treated for 2 years or less. However, we have identified 4 patients that sustained vertebral fractures with discontinuation after just 1 year (two doses) of denosumab treatment. No case has been reported so far with fractures after a single dose of denosumab.

A second concern with denosumab discontinuation is the marked decrease of BMD.⁽², ³⁾ Again, the magnitude of BMD decrease may be linked to the duration of denosumab treatment. In the study of Bone and colleagues,⁽²⁾ women received denosumab for 2 years and then were followed every 6 months for another 2 years. BMD values returned to baseline values 1 year after denosumab discontinuation. On the contrary, women who were treated for a longer period, eg, 10 years, and then discontinued denosumab experienced a decrease in total hip BMD that reached levels lower than baseline (10‐year denosumab gain of 8.2% ± 1.7% versus 1‐year off‐treatment loss of –12.5 ± 1.4%; net loss –5.4% ± 1.8%).⁽¹³⁾ Unfortunately, BMD values 1 year after treatment discontinuation are not available for the patients we studied, mainly because most incidents occurred before the completion of 1 year without treatment.

The risk of fracture after cessation of denosumab was analyzed in the FREEDOM pivotal study. Women who discontinued denosumab after two to five doses reported no excess of fractures during the off‐treatment period.⁽¹¹⁾ However, the median off‐treatment duration was only 8 months, and at least one‐third of the patients had begun an osteoporosis treatment during this period. A more thorough analysis of the data was presented by Brown and colleagues during the 2016 ASBMR Congress.⁽¹⁴⁾ Women who received at least two doses of denosumab and stayed in the study at least 7 months after the last denosumab injection were included (n = 1001). Fifty‐six of them sustained new vertebral fractures (5.6%), a percentage comparable with those who discontinued placebo. However, a greater percentage experienced multiple vertebral fractures in the denosumab group (60.7% versus 34.5%). This analysis has been criticized. First, a certain percentage of women had already begun an osteoporosis treatment during this period. Second, the number of fractures observed in a few months was abnormally high in the placebo group compared with the fracture risk of these patients. Therefore, the actual number of patients who experience fractures after denosumab discontinuation may be higher than 5.6%. It is not possible to be more precise, even by analyzing the practices in our centers. Most of the cases reported were referred to our centers by their GPs to investigate these multiple vertebral fractures. On the other hand, the management of denosumab discontinuation varies from one center to another. This illustrates the need for registries or randomized controlled trials to better understand the risk of multiple vertebral fractures after denosumab discontinuation.

One‐third of the patients we recorded had prevalent vertebral fractures. This may suggest that such patients already have impaired bone strength and, therefore, are prone to new fractures. Prevalent vertebral fractures, before or during the treatment period, were the strongest predictor of new vertebral fractures after discontinuation in the recent analysis of denosumab's pivotal study.⁽¹⁴⁾ These patients should not discontinue denosumab treatment because the presence of fracture outweighs the improvement of BMD. Along with them, treatment should also be continued in patients without prior fractures but whose T‐score values remain in the osteoporotic range. However, we emphasize that most patients were at “low risk” in terms of absence of previous fractures and T‐score values within the osteopenic or even normal range at the time of denosumab discontinuation (Table 1). Therefore, steps should be taken to prevent the rapid rebound in remodeling in most if not all patients who have received at least two doses of denosumab.

Vertebral fractures were multiple in the majority of the patients. Only 2 of 24 had only 1 fracture. This indicates a rapid detrimental effect of denosumab withdrawal that assaults the entire skeleton simultaneously. However, the location of the fractures does not differ from the common osteoporotic vertebral fractures, mostly affecting the lower thoracic and the upper lumbar spine. This is an additional ascertainment that these are typical insufficiency fractures, albeit in a rather magnified scale.

Despite the presence of multiple vertebral fractures, none of the cases reported herein sustained any non‐vertebral fractures. Given its faster remodeling rate, the trabecular bone is more rapidly affected than the cortical bone from the excessively increased bone turnover that ensues from denosumab discontinuation.

All cases occurred 8 to 16 months after the last denosumab injection, therefore, 2 to 10 months after the last dose effect was depleted. Consequently, the first year off‐treatment seems to be the critical period, while after that the risk might be reduced. Because there were incidents as soon as 2 months off‐treatment and given that in the everyday clinical practice patients frequently neglect to receive their next injection on time, physicians should focus more on adherence to treatment and highlight the importance of following the timeline to their patients.

Patient's age is probably of minimal importance because these incidents have been described in a wide range of age. On the other hand, concomitant AI administration may aggravate the withdrawal effect of denosumab on the skeleton, even in normal BMD values. However, because we do not have data on what percentage of women sustained vertebral fractures among those under AIs who discontinued denosumab, this observation remains highly hypothetical.

It had been proposed that these incidents occur in treatment‐naïve patients only.⁽⁸⁾ However, as more cases are reported, it seems that previous treatment might not necessarily avert the risk of fracture. This might be the case for previous anabolic (teriparatide for 1 year, 1 patient), mild antiresorptive treatment (raloxifene for 5 years, 1 patient), or even bisphosphonates when administered for a short period (1 patient) or several years before denosumab initiation (1 patient). Bisphoshonate administration right before or immediately after denosumab treatment has been reported to reduce the rebound of bone turnover markers and potentially diminish the risk of fracture.⁽⁸⁾ After the incident fracture, several treatment strategies were followed: most patients received teriparatide, some reinitiated denosumab, a few received zoledronic acid, and others received a combination of teriparatide and denosumab or zoledronic acid. Although optimal treatment is unknown, it is rational to administer a potent antiosteoporotic agent. Teriparatide may offer pain relief that is more than welcome in these patients with several clinical fractures. However, a point that should be considered in these patients is that teriparatide may increase the already increased bone turnover,⁽¹⁵⁾ which is currently unknown whether it would be beneficial or detrimental, at least at the initial period of treatment. Therefore, it might be safer to administer a potent antiresorptive agent at this setting.

Another clinically significant observation is the failure of vertebroplasty in the few patients that were subjected to it. In all cases, several new fractures occurred in the month after vertebroplasty, questioning the utility of this procedure in these patients. Probably bone strength is seriously affected even in intact vertebrae; therefore, they easily sustain fracture when increased compressing forces are exerted upon them by the neighboring cemented vertebrae. If this is the case, a pharmaceutical agent should be administered before attempting vertebroplasty in order to improve bone strength.

Because of the severity of the rebound‐associated vertebral fractures after denosumab discontinuation, the health authorities, physicians, and patients must be aware of this risk. The Swiss Agency for Therapeutic Products officially recognized the link between denosumab discontinuation and increased risk of multiple vertebral fractures on December 21, 2016 (www.swissmedic.ch/index.html?lang=en).

The major limitation of this review is that it summarizes only case reports or case series, which lie low at the pyramid of evidence. However, to date these studies are the only ones published. Furthermore, the design of a clinical trial oriented to potential bone fragility after discontinuation of denosumab meets ethical considerations, as some patients would remain untreated despite the possibility of an increased fracture risk. Further studies are needed to identify markers (clinical, radiographical, and/or biochemical) for bone fragility after denosumab discontinuation. The most important, however, is to enhance physicians’ awareness of post‐denosumab risk of fragility. Until further data are published, denosumab discontinuation should be carefully considered and, if decided, another anti‐osteoclastic treatment should be administered to avoid bone turnover rebound followed by increased risk of fractures.

Disclosures

Athanasios D. Anastasilakis has received lecture fees from Amgen, Eli‐Lilly, ITF Hellas, ELPEN, VIANEX. Polyzois Makras has received lecture fees and research grants from Amgen; lecture fees from Glaxo, Lilly, Pfizer, Leo, Genesis, ELPEN, VIANEX. Stergios A. Polyzos has received lecture fee from Amgen. Olivier Lamy has received research grants and lecture fees from Amgen, Eli‐Lilly and Takeda.

Acknowledgments

The present study received no funding from any source.

Authors’ roles: Study design: ADA and OL. Study conduct: ADA. Data collection: ADA, SAP, PM, BAR, SK, and OL. Data analysis: ADA and SAP. Data interpretation: ADA and OL. Drafting manuscript: ADA. Revising manuscript content: SAP, PM, BAR, and OL. Approving final version of manuscript: ADA, SAP, PM, BAR, SK, and OL. ADA, SAP, and OL take responsibility for the integrity of the data analysis.